VCF format: indel with two reference alleles

Question

[philip.jonsson]

I'm struggling with making VarScan's output adhere to GATK's strict rules for VCF files. Currently, I cannot understand what is wrong with this line:

1    20571284    .    CAAAAAAA,AAAAAAAA    C    .

GATK says this cannot be parsed. Are two reference alleles not allowed, or is is supposed to be formatted otherwise?

[lh3]

EDIT: OK, what is the reference base at 20571284? Is it C or A or M or N?

Answer 1

[Pierre Lindenbaum]

No the VCF spec says that there is only one REF allele: http://samtools.github.io/hts-specs/VCFv4.2.pdf while you can have one or more ALT bases

REF - reference base(s): Each base must be one of A,C,G,T,N (case insensitive). Multiple bases are permitted.

ALT: Comma separated list of alternate non-reference alleles called on at least one of the samples.

to solve your problem, you can duplicate your VCF line:

1    20571284    .    AAAAAAAA    C    .
1    20571284    .    CAAAAAAA    C    .

but (as said Heng below) you'll also have to check the following statement of the spec: (REF) "Strings must include the base before the event"

[lh3]

These are two different REF alleles, which cannot be right.

[philip.jonsson]

Alright, thank you. I suspected that was the case but didn't find anything that verified it.